{
  "best_metric": 0.8967889908256881,
  "best_model_checkpoint": "electra-distilled-sst\\run-28\\checkpoint-527",
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 527,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 33.01731872558594,
      "learning_rate": 3.176988118911755e-05,
      "loss": 2.1443,
      "step": 527
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8967889908256881,
      "eval_loss": 1.0754104852676392,
      "eval_runtime": 1.4504,
      "eval_samples_per_second": 601.214,
      "eval_steps_per_second": 4.826,
      "step": 527
    }
  ],
  "logging_steps": 500,
  "max_steps": 4216,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 8,
  "save_steps": 500,
  "total_flos": 186587234208876.0,
  "train_batch_size": 128,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.5452558462704341,
    "learning_rate": 3.628876160751005e-05,
    "num_train_epochs": 8,
    "temperature": 24
  }
}